A Novel Approach for Optimizing Wire Electric Discharge Machining of Mg-Cu-RE-Zr Alloy Using Machine Learning Algorithm

Abstract

This study focuses on the optimisation of the wire electric discharge machining (WEDM) process for WE43 alloy using machine learning methods. The alloy, made of magnesium (Mg), copper (Cu), rare earth (RE) elements, and zirconium (Zr), is extensively employed in aerospace and automotive sectors for its lightweight and high-strength features. The research applies three machine learning models—artificial neural networks (ANN), random forest (RF), and decision trees (DT)—to optimize the important process parameters, including current (A), pulse on (P On), and pulse off (P Off). A full experimental design based on the Taguchi L27 array is undertaken, methodically altering each parameter at three levels. Material removal rate (MRR) is chosen as the response variable for optimisation. The process parameters are adjusted by the use of machine learning techniques, with ANN emerging as the most accurate predictor, obtaining an accuracy of 96.7%.